Latest Gold Price
4345.93 USD
TIPS: 2.68% | CN_US_Spread: -3.25%
Gold-Oil Ratio: 45.1
Valuation Status
1-Year Percentile 65.9% Fair
Full History Percentile 96.7% Extremely Overvalued
📈 Multi-Window Valuation Percentile Table (since 2016)
| Window | Percentile | Status |
|---|---|---|
| 1-Year | 65.9% | Fair |
| 3-Year | 88.6% | Overvalued |
| 5-Year | 93.0% | Extremely Overvalued |
| 10-Year | 96.5% | Extremely Overvalued |
| Full History | 96.7% | Extremely Overvalued |
💡 How to read: Higher percentile means gold is more expensive relative to the model-implied equilibrium. Full history percentile reflects long-cycle temperature; 1-year percentile captures short-term overheating.
Zone thresholds: >90% (Extremely Overvalued) | 70%-90% (Overvalued) | 30%-70% (Fair) | 10%-30% (Undervalued) | <10% (Extremely Undervalued).
📉 Strategy Backtest & Performance
Valuation-percentile-based strategies have performed well in historical backtests. All strategies use the previous day's valuation percentile to determine today's position, avoiding look-ahead bias.
Strategy Rules
| Strategy | Rebalancing Rule |
|---|---|
| Buy and Hold | Always fully invested in gold |
| Linear Position (Full History) | Position = 1 - full history percentile |
| Extreme Timing (Extreme Reversal) | Fully long when Extremely Undervalued, zero when Extremely Overvalued, otherwise unchanged |
| Smart Position (1-Year) | Piecewise: q<0.1→3-5q, 0.1~0.3→2.5-5(q-0.1), 0.3~0.7→1.5-2.5(q-0.3), 0.7~0.9→0.5-2(q-0.7), >0.9→0.1 |
| Trend-Cautious Fusion | Smart Position basis, but position halved when price below 200-day moving average |
Performance (2016-04-18 to 2026-09-18)
| Strategy | Total Return | Annual Return | Sharpe Ratio | Max Drawdown |
|---|---|---|---|---|
| Buy and Hold | 252.5% | 12.5% | 0.79 | -26.6% |
| Linear Position (Full History) | 253.5% | 12.5% | 1.42 | -9.8% |
| Extreme Timing (Extreme Reversal) | 292.1% | 13.6% | 1.11 | -18.4% |
| Smart Position (1-Year) | 1993.4% | 32.9% | 1.41 | -22.6% |
| Trend-Cautious Fusion | 2407.2% | 35.1% | 1.62 | -20.4% |
📌 Strategy Performance Interpretation:
- Buy and Hold serves as the benchmark, showing gold's inherent long-term return.
- Linear Position reduces volatility and drawdowns effectively through the simple "more expensive → lighter position" mechanism, sharply improving the Sharpe ratio.
- Extreme Timing frequently misses rallies in trending bull markets, indicating the risk of relying solely on extreme thresholds.
- Smart Position uses an asymmetric piecewise function to moderately leverage when undervalued and heavily cut positions when overvalued, achieving the highest absolute return.
- Trend-Cautious Fusion adds a trend filter to the Smart Position, delivering the best risk-adjusted return with well-controlled drawdowns—the most practically relevant approach.
📌 Full-cycle robustness: If the backtest is extended to 2006–2026 (including the 2008 financial crisis and the 2011–2015 gold bear market), Smart Position has an annual return of 21.8%, Sharpe 0.78, max drawdown -57.9%, still significantly outperforming Buy and Hold (annual 9.92%, Sharpe 0.56).
Note: Backtests assume zero transaction costs and no financing interest on leverage. Past performance does not guarantee future results. Leverage amplifies losses; practical application requires caution.
📊 Gold vs. Factor Relationships
📌 Chart 1: Gold has risen from about $1,200 to 4346 over the past decade. The current 3-year rolling percentile is 88.6%, in the Extremely Overvalued zone. Central bank gold purchases recently around 55.0 tons/month, providing structural support.
📌 Chart 2: TIPS currently 2.68%, China-U.S. yield spread -3.25%. The gold-oil ratio is 45.1, reflecting how many barrels of WTI one ounce of gold can buy. A rising ratio usually signals heightened risk aversion or divergent inflation expectations.
📊 Residual Trend
📌 Chart 3: Current residual is 0.1879, in the Extremely Overvalued zone. A persistently positive residual suggests gold is driven by structural forces beyond the factors. High residual phases often coincide with extreme market sentiment or geopolitical risk premiums.
🧠 Model Interpretation
Regression Equation: Ln(Gold) = 7.3936 + -0.3007×TIPS + -31.5660×CN_US_Spread + 0.0110×Gold_Oil_Ratio, R²=0.912
Factor Explanations:
- TIPS (U.S. 10-Year Real Yield): Measures the opportunity cost of holding gold. Rising TIPS → gold under pressure; negative coefficient aligns with classical pricing theory.
- China-U.S. Yield Spread (CN_US_Spread): China 10Y minus U.S. 10Y, capturing global capital allocation between China and the U.S. When the spread narrows/inverts, capital may shift to gold as a substitute, pushing prices higher.
- Gold-Oil Ratio (Gold_Oil_Ratio): The ratio of gold to WTI crude oil, reflecting risk aversion and relative commodity valuation. A rising ratio usually signals increasing safe-haven demand.
Bubble Duration:Extremely Overvalued appeared 63 times, avg 11 days, max 117 days
Residual Percentile: Current residual is 0.1879, in the Extremely Overvalued zone. Residual > 0 indicates the price is above the factor-implied equilibrium.
🔬 Collinearity Diagnostics (VIF)
| Variable | VIF | Judgment |
|---|---|---|
| TIPS | 5.49 | ⚠️ |
| CN_US_Spread | 3.56 | ✅ |
| Gold_Oil_Ratio | 2.36 | ✅ |
Note: The TIPS factor's VIF is slightly above 5 (5.01) in some data update windows, indicating mild collinearity. Given that its coefficient sign is consistent with economic theory (negative: TIPS up → gold price under pressure) and it remains marginally significant after Newey-West correction (p=0.051), we believe its independent contribution still deserves inclusion in the model. Moreover, minor VIF fluctuations are associated with format switching (CSV/XLSX) of FRED data sources and do not alter the core conclusions of the model.
Rolling Window Robustness: 2-Year: 82.9% | 3-Year: 88.6% | 5-Year: 93.0%
📊 Multi-Period Comparison
| Period | Price Change | 3Y Percentile Change | Zone Change |
|---|---|---|---|
| Week | 4348.35→4345.93 (-0.1%) | 90.2%→88.6% (-1.6%) | Extremely Overvalued→Extremely Overvalued |
| Month | 4522.78→4345.93 (-3.9%) | 88.8%→88.6% (-0.1%) | Extremely Overvalued→Extremely Overvalued |
| Quarter | 4155.44→4345.93 (+4.6%) | 90.1%→88.6% (-1.5%) | Extremely Overvalued→Extremely Overvalued |
| Half Year | 4496.98→4345.93 (-3.4%) | 99.3%→88.6% (-10.7%) | Extremely Overvalued→Extremely Overvalued |
| Year | 3644.27→4345.93 (+19.3%) | 42.9%→88.6% (+45.8%) | Fair→Extremely Overvalued |
📎 Key Levels: YTD 3Y percentile high 100.0% | low 19.8%; gold price one year ago 3644.27 (YoY +19.3%).
📋 Forward Return Statistics
| Zone | Samples | 1-Month | 3-Month | 6-Month |
|---|---|---|---|---|
| Extremely Undervalued | 394 | 7.3% | 24.2% | 52.6% |
| Undervalued | 426 | 7.5% | 24.0% | 45.0% |
| Fair | 783 | 4.1% | 12.9% | 25.3% |
| Overvalued | 589 | 6.1% | 15.3% | 28.1% |
| Extremely Overvalued | 504 | 8.5% | 27.8% | 63.9% |
⏱️ Valuation Zone Duration Statistics (based on full history percentile)
Currently in Extremely Overvalued state, lasting 33 days. Current full history percentile is 96.7%, within the 90%-100% band, which has lasted 44 days.
| Valuation Zone | Occurrences | Avg Duration | Min | Max | Median Duration |
|---|---|---|---|---|---|
| Extremely Undervalued | 53 times | 7.4 d | 1 d | 58 d | 3.0 d |
| Undervalued | 150 times | 2.8 d | 1 d | 23 d | 2.0 d |
| Fair | 185 times | 4.2 d | 1 d | 27 d | 2.0 d |
| Overvalued | 129 times | 4.6 d | 1 d | 60 d | 2.0 d |
| Extremely Overvalued | 65 times | 7.8 d | 1 d | 82 d | 2.0 d |
📅 Current Percentile Band Duration
| Percentile Band | Occurrences | Avg Duration | Min | Max | Median Duration |
|---|---|---|---|---|---|
| 90%-100% (current) | 76 times | 8.6 d | 1 d | 79 d | 3.5 d |
Tactical Reference: For long-term allocation, wait for full history percentile to drop below 70%; for short-term trading, focus on the 1-year percentile; if the gold-oil ratio surges, be alert to extreme risk aversion.
🕒 Factor Data Freshness
Valuation date is 2026-09-18 00:00:00. Some macro factors have normal lags; the model applies forward fill.
| Factor | Latest Data Date | Status |
|---|---|---|
| London Gold (XAU) | 2026-09-18 | ✅ Real-time |
| TIPS Real Yield | 2026-09-16 | ⚠️ Lag 2-3 days |
| Shanghai Gold AU9999 | 2026-09-15 | ⚠️ Lag 2-3 days |
| USD/CNY | 2026-09-17 | ✅ Real-time |
| China 10-Year Bond | 2026-09-17 | ✅ Real-time |
| U.S. 10-Year Bond | 2026-09-17 | ✅ Real-time |
| Central Bank Buying | 2026-07-31 | 📅 Lag 49 days |
| Gold-Oil Ratio | 2026-09-18 | ✅ Real-time |